01/09/2012 12:49 PMPosted by HalloomThe fairly small DPS display difference can be easily accounted for by the DPS calculations being per-item, for ease of recalculation when swapping gear, and simply rounding off at any point between them.

your conclusion is based on a hypothesis of two items, but did you bother to confirm the testable predictions posited; comparison of single items? /rhetorical

1a) Dps calc is accurate to the 100th decimal, a difference at that point

*is*acceptable, and can be demonstrated often

1b) Discrepancy of 2/12 (16.66%) is

*not*

2) A rounding issue of this magnitude would be ubiquitous, this is much more selective*

-Testable prediction-

A) Given a single rare & magical ring with

*identical*stats, from seperate item calculations the dps must

*not*be off (

*give blizz more credit than that, it wouldn't be that wrong*)."

B)

*continued*Given a single rare & magical ring with

*unique*stats, more powerful stats

*may*trigger higher than expected, conversely weaker stats lower - but if rounding did clip this much, it cannot also create more when total stats are lower.

confirmation of B

http://oi41.tinypic.com/nf0bbc.jpg

http://oi42.tinypic.com/ajotg2.jpg

magical (blue) with a full IAS more, is somehow rounding 2.15% lower, thus I reject your Hypothesis. This is called "peer review." methodology, prediction, evidence...

Learn science.

what I like is how clearly the comparison text is telling us that rare missing 1 ias over the blue is a full +7dps , and *the mod that these examples have in common is 3-8

"of wounding" suffix. mmm a good mystery!

^this

why doesn't everyone get it?

^this

why doesn't everyone get it?

Probably because that isn't the issue. We all understand that an increase from 250% increased attack damage to 254% will not be perceived as an overall increase of 4% dps. That isn't the issue at all, however.

This seems to be a legitimate problem. I have no explanation for why seemingly identical stats can result in different DPS just because the items are swapped out. It honestly seems like a bug with the DPS calculation to me, which could very likely also result in a bug with realized DPS in combat.

Unless someone can come up with a reasonable explanation, a blue needs to be aware of this.

http://us.battle.net/d3/en/forum/topic/3840666169

bug, or hidden stats, we just can''t figure out a way to deduce which though.

Because my ring that has 2-9 damage deals more damage than my 3-11 ring.

There's no possible way right? It's just damage calculation errors, or hidden stats that's bugged.

Hopefully the developers sees this and attempts to fix it.

01/09/2012 10:06 PMPosted by D3BETAyour conclusion is based on a hypothesis of two items, but did you bother to confirm the testable predictions posited; comparison of single items? /rhetorical

No. You're welcome to actually test

*my actual statement*- but I suppose you couldn't if you never understood it.

Calculate the damage dealt by the character per-item, rounding the

*damage dealt*to the nearest whole number (or, more likely, tenth) after each calculated item, then derive the DPS from this, for each scenario.

Learn

*computer*science.

what I like is how clearly the comparison text is telling us that rare missing 1 ias over the blue is a full +7dps , and *the mod that these examples have in common is 3-8

"of wounding" suffix. mmm a good mystery!

It's not a mystery if you know how computers work.

'Floating point' mathematics, which deals with percentages at a high degree of precision, is massively computationally wasteful.

So wasteful, in fact, that even when you're

*literally dealing with percentages*, sometimes you don't want to use it.

An easy way to introduce fake precision while still using efficient integer math is to simply use integers and pretend that you're working in fractions - calculating 10.1 as 101 behind the scenes and just operating as if it expresses the value in a quantity of tenths, for example.

The downside of this efficient math is that integer math has

*no additional precision at all*, and so every distinct operation you perform on it has a likelihood of introducing a rounding error. Enough steps - and enough rounding errors - and you can get something visible even if you're using integer math at a 'fake precision' level 1 magnitude higher than the precision the user can see.

Unless and until yELo would care to share details of how

*his*damage rings are working (Edit: I read yELo's post detailing his rings. It's not a bug either, he just doesn't think +crit damage increases your damage), that's almost certainly what's going on. A fraction of a DPS difference is easily consistent with a rounding error to the nearest tenth, or even a cumulative series of rounding errors to the nearest hundredth. Furthermore, a small rounding error in a multiplicative field - such as attack speed - can be magnified significantly (We'll see if you actually read this far, eh?) because you're then multiplying the number.

Really, this entire thread is a homage to how arcane mathematics are to normal people, and to how arcane mathematics computers do is to normal mathematicians.

us.battle.net/d3/en/forum/topic/3840666169#10

still blabbing arrogantly from a position of ignorance halloom?... some baddie trolling if I ever saw it.

No because it is a straw man, I won't do your homework testing negative arguments; there is no evidence to support the hypothesis.

You're welcome to actually testmy actual statement

First demonstrate that you can reproduce the issue, that is called

*the burden of proof*....

then i'll consider rounding the culprit,

But you can not/or will not

I'll quote myself again fo the emphasis...

01/09/2012 10:06 PMPosted by D3BETAA rounding issue of this magnitude would be ubiquitous

follow the evidence

http://us.battle.net/d3/en/forum/topic/3840666169#10